Hyperspectral Inversion Of Soil Nutrients In Mining Area And Its Influence Analysis

While coal mining brings economic growth,it also causes surface subsidence and collapse,and the surface is prone to form fissures,resulting in the destruction of soil properties and types,as well as the loss of water and soil,nutrients,etc.,which threatens the safety of villagers around the mining area,and the contradiction between human and land is sharpening,soil management and restoration has become an important part of mine restoration,of which,soil physical and chemical property monitoring is an important part.The development and application of hyperspectral technology has provided new methods and ideas for soil nutrient and physicochemical property monitoring.In this study,based on the summary and analysis of the technical methods of hyperspectral monitoring of soil nutrients,the sensitivity of different spectral bands to soil nutrients was compared and analyzed by using field sampling and indoor hyperspectral measurement in the experimental area of Zhuzhuang mining area in Huaibei city,and the characteristic bands of hyperspectral monitoring of soil nutrients were studied,and based on this,we compare and analyze BP neural network(BPNN,partial least squares(PLSR,support vector machine(SVM and other soil nutrient inversion models and methods,and study the best inversion model for soil nutrient hyperspectral;using the distance between the sampling point and the fissure as a reference,we analyze the different soil nutrient changes with the increase of the distance from the fissure,and the factors affecting the soil nutrient properties at the fissure.The main contents and conclusions are as follows:1 Analysis and selection of the characteristic spectral bands of soil nutrients.Based on different spectral transformations: smoothed spectral reflectance(R,first-order differential(FDR,second-order differential(SDR,absorbance transformation(Log(1/ R)and soil nutrients in organic matter,PH,effective phosphorus and fast-acting potassium soil attributes,the bands with significant significance test at p(28)0.01 level were screened as the characteristic bands and the characteristic soil nutrient attributes were determined waveform.2 Analysis and establishment of the best inversion model for different soil attributes.After comparing and analyzing the inversion models of four soil attributes,the best model for organic matter was a BP neural network model based on first-order differentiation,the best model for PH was a partial least squares model with first-order differentiation,the best model for effective phosphorus was a partial least squares model with absorbance transformation,and the best model for fast-acting potassium was a support vector machine model with first-order differentiation.3 Analysis of soil property changes on both sides of the fissure.By comparing the changes in soil properties at different distances from the fissure,it can be seen that the PH value decreases with the increase in the distance from the fissure,and the values of organic matter,effective phosphorus and fast-acting potassium at the nearest and farthest points from the fissure are much higher than those at other sampling points,and the values of properties at the farthest point from the fissure are greater than the values closer to the fissure.4 Combining the remote sensing image data and meteorological station data of different time periods,the influence of vegetation and precipitation on the changes of nutrient properties at soil fissures was studied and analyzed.The results show that organic matter and fast-acting potassium attributes are more influenced by vegetation,effective phosphorus attributes are more influenced by precipitation,and PH attributes are less influenced by precipitation and vegetation with less changes during the study period.Figure.[22] table [10] Reference.[84]...